What it is
AI-Trading is a Python experiment in LLM-driven portfolio allocation. It starts with a virtual 10,000 USDT balance, fetches current market data, asks an AI model for target allocations, and records the resulting simulated portfolio over time.
How it works
The bot pulls 15-minute market data for BTC, ETH, and SOL through yfinance, sends the market context to DeepSeek through OpenRouter, normalizes the model response into valid JSON targets, converts those targets into crypto quantities, and appends each run to a CSV log.
What shipped
- Paper-trading loop with no real funds at risk.
- OpenRouter model call for portfolio target allocation.
- Scheduler support for repeated runs every 16 minutes.
- Persistent CSV log for tracking decisions and portfolio movement.
Important note
This is a simulation, not financial advice, and it does not place real trades. The interesting part is the control loop: live data, model decision, normalized allocation, and repeatable logging.